The Fermi-Dirac distribution provides a calibrated probabilistic output for binary classifiers.

Binary classification is one of the central problems in machine-learning research and, as such, investigations of its general statistical properties are of interest. We studied the ranking statistics of items in binary classification problems and observed that there is a formal and surprising relationship between the probability of a sample belonging to one of the two classes and the Fermi-Dirac distribution determining the probability that a fermion occupies a given single-particle quantum state in a physical system of noninteracting fermions. Using this equivalence, it is possible to compute a calibrated probabilistic output for binary classifiers. We show that the area under the receiver operating characteristics curve (AUC) in a classification problem is related to the temperature of an equivalent physical system. In a similar manner, the optimal decision threshold between the two classes is associated with the chemical potential of an equivalent physical system. Using our framework, we also derive a closed-form expression to calculate the variance for the AUC of a classifier. Finally, we introduce FiDEL (Fermi-Dirac-based ensemble learning), an ensemble learning algorithm that uses the calibrated nature of the classifier's output probability to combine possibly very different classifiers.

Dynamical processes of interstitial diffusion in a two-dimensional colloidal crystal.

In two-dimensional (2D) solids, point defects, i.e., vacancies and interstitials, are bound states of topological defects of edge dislocations and disclinations. They are expected to play an important role in the thermodynamics of the system. Yet very little is known about the detailed dynamical processes of these defects. Two-dimensional colloidal crystals of submicrometer microspheres provide a convenient model solid system in which the microscopic dynamics of these defects can be studied in real time using video microscopy. Here we report a study of the dynamical processes of interstitials in a 2D colloidal crystal. The diffusion constants of both mono- and diinterstitials are measured and found to be significantly larger than those of vacancies. Diinterstitials are clearly slower than monointerstitials. We found that, by plotting the accumulative positions of five- and sevenfold disclinations relative to the center-of-mass position of the defect, a sixfold symmetric pattern emerges for monointerstitials. This is indicative of an equilibrium behavior that satisfies local detailed balance that the lattice remains elastic and can be thermally excited between lattice configurations reversibly. However, for diinterstitials the sixfold symmetry is not observed in the same time window, and the local lattice distortions are too severe to recover quickly. This observation suggests a possible route to creating local melting of a lattice (similarly one can create local melting by creating divacancies). This work opens up an avenue for microscopic studies of the dynamics of melting in colloidal model systems.

Unannotated small RNA clusters associated with circulating extracellular vesicles detect early stage liver cancer.

OBJECTIVE: Surveillance tools for early cancer detection are suboptimal, including hepatocellular carcinoma (HCC), and biomarkers are urgently needed. Extracellular vesicles (EVs) have gained increasing scientific interest due to their involvement in tumour initiation and metastasis; however, most extracellular RNA (exRNA) blood-based biomarker studies are limited to annotated genomic regions. DESIGN: EVs were isolated with differential ultracentrifugation and integrated nanoscale deterministic lateral displacement arrays (nanoDLD) and quality assessed by electron microscopy, immunoblotting, nanoparticle tracking and deconvolution analysis. Genome-wide sequencing of the largely unexplored small exRNA landscape, including unannotated transcripts, identified and reproducibly quantified small RNA clusters (smRCs). Their key genomic features were delineated across biospecimens and EV isolation techniques in prostate cancer and HCC. Three independent exRNA cancer datasets with a total of 479 samples from 375 patients, including longitudinal samples, were used for this study. RESULTS: ExRNA smRCs were dominated by uncharacterised, unannotated small RNA with a consensus sequence of 20 nt. An unannotated 3-smRC signature was significantly overexpressed in plasma exRNA of patients with HCC (p<0.01, n=157). An independent validation in a phase 2 biomarker case-control study revealed 86% sensitivity and 91% specificity for the detection of early HCC from controls at risk (n=209) (area under the receiver operating curve (AUC): 0.87). The 3-smRC signature was independent of alpha-fetoprotein (p<0.0001) and a composite model yielded an increased AUC of 0.93. CONCLUSION: These findings directly lead to the prospect of a minimally invasive, blood-only, operator-independent clinical tool for HCC surveillance, thus highlighting the potential of unannotated smRCs for biomarker research in cancer.

Broken flow symmetry explains the dynamics of small particles in deterministic lateral displacement arrays.

Deterministic lateral displacement (DLD) is a technique for size fractionation of particles in continuous flow that has shown great potential for biological applications. Several theoretical models have been proposed, but experimental evidence has demonstrated that a rich class of intermediate migration behavior exists, which is not predicted. We present a unified theoretical framework to infer the path of particles in the whole array on the basis of trajectories in a unit cell. This framework explains many of the unexpected particle trajectories reported and can be used to design arrays for even nanoscale particle fractionation. We performed experiments that verify these predictions and used our model to develop a condenser array that achieves full particle separation with a single fluidic input.

Modeling combination therapies in patient cohorts and cell cultures using correlated drug action.

Characterizing the effect of combination therapies is vital for treating diseases like cancer. We introduce correlated drug action (CDA), a baseline model for the study of drug combinations in both cell cultures and patient populations, which assumes that the efficacy of drugs in a combination may be correlated. We apply temporal CDA (tCDA) to clinical trial data, and demonstrate the utility of this approach in identifying possible synergistic combinations and others that can be explained in terms of monotherapies. Using MCF7 cell line data, we assess combinations with dose CDA (dCDA), a model that generalizes other proposed models (e.g., Bliss response-additivity, the dose equivalence principle), and introduce Excess over CDA (EOCDA), a new metric for identifying possible synergistic combinations in cell culture.

A mask R-CNN based automatic assessment system for nail psoriasis severity.

Nail psoriasis significantly impacts the quality of life in patients with psoriasis, which affects approximately 2-3% of the population worldwide. Disease severity measures are essential in guiding treatment and evaluation of therapeutic efficacy. However, due to subsidy, convenience and low costs of health care in Taiwan, doctor usually needs to manage nearly hundreds of patients in single outpatient clinic, leading to difficulty in performing complex assessment tools. For instance, Nail Psoriasis Severity index (NAPSI) is used by dermatologists to measure the severity of nail psoriasis in clinical trials, but its calculation is quite time-consuming, which hampers its application in daily clinical practice. Therefore, we developed a simple, fast and automatic system for the assessment of nail psoriasis severity by constructing a standard photography capturing system combined with utilizing one of the deep learning architectures, mask R-CNN. This system not only assist doctors in capturing signs of disease and normal skin, but also able to extract features without pre-processing of image data. Expectantly, the system could help dermatologists make accurate diagnosis, assessment as well as provide precise treatment decision more efficiently.

Deterministic Lateral Displacement: Challenges and Perspectives.

The advent of microfluidics in the 1990s promised a revolution in multiple industries from healthcare to chemical processing. Deterministic lateral displacement (DLD) is a continuous-flow microfluidic particle separation method discovered in 2004 that has been applied successfully and widely to the separation of blood cells, yeast, spores, bacteria, viruses, DNA, droplets, and more. Deterministic lateral displacement is conceptually simple and can deliver consistent performance over a wide range of flow rates and particle concentrations. Despite wide use and in-depth study, DLD has not yet been fully elucidated or optimized, with different approaches to the same problem yielding varying results. We endeavor here to provide up-to-date expert opinion on the state-of-art and current fundamental, practical, and commercial challenges with DLD as well as describe experimental and modeling opportunities. Because these challenges and opportunities arise from constraints on hydrodynamics, fabrication, and operation at the micro- and nanoscale, we expect this Perspective to serve as a guide for the broader micro- and nanofluidic community to identify and to address open questions in the field.

Gel-on-a-chip: continuous, velocity-dependent DNA separation using nanoscale lateral displacement.

We studied the trajectories of polymers being advected while diffusing in a pressure driven flow along a periodic pillar nanostructure known as nanoscale deterministic lateral displacement (nanoDLD) array. We found that polymers follow different trajectories depending on their length, flow velocity and pillar array geometry, demonstrating that nanoDLD devices can be used as a continuous polymer fractionation tool. As a model system, we used double-stranded DNA (dsDNA) with various contour lengths and demonstrated that dsDNA in the range of 100-10 000 base pairs (bp) can be separated with a size-selective resolution of 200 bp. In contrast to spherical colloids, a polymer elongates by shear flow and the angle of polymer trajectories with respect to the mean flow direction decreases as the mean flow velocity increases. We developed a phenomenological model that explains the qualitative dependence of the polymer trajectories on the gap size and on the flow velocity. Using this model, we found the optimal separation conditions for dsDNA of different sizes and demonstrated the separation and extraction of dsDNA fragments with over 75% recovery and 3-fold concentration. Importantly, this velocity dependence provides a means of fine-tuning the separation efficiency and resolution, independent of the nanoDLD pillar geometry.